1. Field of the Invention
The present invention relates to an image processing apparatus and, more particularly, to an image processing apparatus for improving quality of a multi-generation copy.
2. Related Background Art
A digital copying machine is known well as an apparatus for causing a reader to read an original, converting a read image into an electrical signal, and driving a record head or a laser to form an image in accordance with this electrical signal, thereby forming a copy. This digital copying machine has features such as image processing and image editing and has been very popular.
A digital color copying apparatus is superior to an analog color copying apparatus in image quality such as color reproducibility. Because of the possibility of obtaining various types of color copies by utilizing various editing functions, the market in this field is rapidly expanding.
FIG. 2 is a block diagram showing a flow of an image signal in the above digital color copying apparatus.
Referring to FIG. 2, a CCD 1 outputs R (red), G (green), and B (blue) signals 2a, 2b, and 2c to a black offset & shading correction circuit 3. The black offset & shading correction circuit 3 outputs shading-corrected R, G, and B signals 4a, 4b, and 4c to a log conversion circuit 5. The log conversion circuit 5 outputs C (cyan), M (magenta), and Y (yellow) density signals 6a, 6b, and 6c to a color processing circuit 7. The color processing circuit 7 outputs color-processed C, M, Y, and Bk (black) signals 8a, 8b, 8c, and 8d to a gamma correction circuit 9. The gamma correction circuit 9 outputs gamma-corrected C, M, Y, and Bk signals 10a, 10b, 10c, and 10d to an edge emphasis & smoothing circuit 20. The edge emphasis & smoothing circuit 20 outputs edge-emphasized and smoothed image signals 21a, 21b, 21c, and 21d to record heads 11a to 11d. The record heads 11a to 11d serve as cyan, magenta, yellow, and black record heads, respectively.
An original image is read by the CCD 1 and converted into electrical signals. The electrical signals are converted into the digital R, G, and B signals 2a, 2b, and 2c by an A/D converter (not shown). These signals are subjected to black offset processing and shading correction processing for correcting variations in the CCD 1 and an original illumination lamp (not shown) by the black offset & shading correction circuit 3.
A standard black board and a standard white board (neither is shown) are arranged in an original reader. By using these boards, the above processing operations are performed. More specifically, the standard black board is a black board having an optical density of 2.0, and the standard white board is a white board having an optical density of 0.07. Values A and B obtained upon reading of the black and white boards are stored in units of pixels. A value X obtained upon reading of an original is converted into the following value by the black offset & shading correction circuit: ##EQU1## In this case, each signal is an 8-bit signal which has a maximum value of 255.
The converted R, G, and B signals are logarithmically converted into the C, M, and Y density signals 6a, 6b, and 6c. These signals are subjected to color correction processing in the color processing circuit 7. Black extraction processing and masking processing are performed in this color correction processing. The black extraction processing is processing for extracting a black component from the C, M, and Y signals to improve reproducibility of black. The masking processing is processing for correcting transmission characteristics of filters within the CCD 1 and reflecting characteristics of inks C, M, Y, and Bk.
If signals input to the color processing circuit 7 are given as C, M, and Y, the black component is extracted by the following calculation in black extraction processing: EQU Bk=min(C, M, Y)
Subsequently, color correction is performed in the masking processing as follows: EQU C'=a.sub.11 C+a.sub.12 M+a.sub.13 Y+a.sub.14 Bk EQU M'=a.sub.21 C+a.sub.22 M+a.sub.23 Y+a.sub.24 Bk EQU Y'=a.sub.31 C+a.sub.32 M+a.sub.33 Y+a.sub.34 Bk EQU Bk'=a.sub.41 C+a.sub.42 M+a.sub.43 Y+a.sub.44 Bk
The masking parameters a.sub.11 to a.sub.44 are set to obtain optimal color reproducibility.
The color-corrected signals (8a to 8d) are gamma-corrected in the gamma correction circuit 9. In the gamma correction circuit 9, gradation characteristics of the heads 11a to 11d are corrected to obtain a Linear relationship between the density signals and the print densities, and at the same time, the C, M, Y, and Bk components are balanced. When the gradation characteristics of the heads are linear, the gamma correction circuit 9 corrects the input C, M, Y, and Bk signals as follows: EQU C'=a.sub.5 .times.C EQU M'=a.sub.6 .times.M EQU Y'=a.sub.7 .times.Y EQU Bk'=a.sub.8 .times.Bk
The gamma-corrected signals 10a to 10d are subjected to edge emphasis and smoothing processing in the edge emphasis & smoothing circuit 20. The edge emphasis and smoothing circuit 20 performs known edge emphasis and known smoothing processing in the following manner.
FIG. 3 is a detailed block diagram of the edge emphasis & smoothing circuit 20. The edge emphasis & smoothing circuit 20 comprises a smoothing unit 22 for receiving the gamma-corrected image signal C on line 10a and outputting a smoothing signal on line 23, a subtracter 24 for outputting an edge component signal on line 25, multipliers 26, 27, and 28 for outputting product output signals on line 29, 30, and 31, respectively, and an adder 32 for outputting an output signal C on line 21a as an output signal from the edge emphasis & smoothing circuit 20.
The smoothing unit 22 averages image signals of neighboring pixels of a target pixel and outputs a smoothing signal S. If a smoothing matrix has a size of 3.times.3 and an image signal of a target pixel position (i,j) is given as f(i,j), the smoothing signal S is calculated as follows: ##EQU2##
The subtracter 24 subtracts the smoothing signal S from the signal f(i,j) of the target pixel to extract an edge component. An edge component signal E is given as follows: EQU E=f(i,j)-S
The multipliers 26, 27, and 28 multiply the target pixel signal, the smoothing signal, and the edge component signal with predetermined coefficients k.sub.1 to k.sub.3, respectively. The adder 32 adds the product components from the multipliers 26, 27, and 28 and outputs an output signal from the edge emphasis & smoothing circuit 20.
The component coefficients k.sub.1, k.sub.2, and k.sub.3 are determined by spatial frequency characteristics and the design concept of the copying machine. For example, when the spatial frequency characteristics of the copying machine are poor and a thin line is blurred, the smoothing signal coefficient k.sub.2 is set to zero, and the edge component signal coefficient k.sub.3 is increased. In order to emphasize a character original and a thin portion with good reproducibility, similar changes are performed. In order to emphasize a picture or the like of a print to obtain a good halftone image without forming a moire pattern, the coefficient k.sub.3 is set to be small, and the coefficient k.sub.2 is increased.
In this case, k.sub.1 +k.sub.2 is kept constant because the sum k.sub.1 +k.sub.2 represents the magnitude of the signal. If this sum is changed, an image density is changed.
The edge emphasis and the smoothing processing are performed for the remaining color signals, i.e., M, Y, and Bk by using common coefficients. The record heads 11a to 11d are driven by the resultant signals, and a color copy is obtained. Each record head comprises an ink-jet head, a thermal head, or the like. When an electrophotographic system is employed, the record head is constituted by a semiconductor laser, an LED array, or a liquid crystal shutter array.
Even if the coefficients k.sub.1, k.sub.2, and k.sub.3 are optimized, and when generation copying, i.e., copying of a copy output (i.e., the first reproduced image upon processing of the original) using as an original, a problem often occurs in reproducibility of an edge portion.
For example, when spatial frequency characteristics of the original copy are poor and correction is to be performed by edge emphasis, or when edge emphasis is performed to improve reproducibility of characters or thin lines, characters and lines are reproduced to be thick. In the second generation copying (an output from a copy original obtained from an original) using a copy output as an original, a thin line is further emphasized to further increase a line width.
When multi-generation copying is repeated, characters and thin lines become contiguous, and readability of the original is impaired, resulting in inconvenience.
In an apparatus which is set in an overemphasized smoothing mode to obtain a smoother halftone image, when generation copying is repeated, character and thin line elements are blurred, and readability of the original is impaired, resulting in inconvenience.
When a copy is to be obtained using an output color copy obtained as described above, a copied product has an excessively low density in a conventional arrangement, thus degrading readability of the copy due to the following reason.
Most copy originals are generally printed matter, and the maximum density of these originals reaches 1.8 to 2.0. Image processing is performed such that an image signal obtained upon reading of a portion having a maximum density of 2.0 is converted into the maximum value 225, of the image signal. When a given density corresponding to 255 represented by an image signal is 2.0 or less, gradation levels of densities higher than the given density cause contiguity of character elements, i.e., cause the character elements to run into each other where they should not. To the contrary, when a density exceeds the given density, all values to be represented by each 8-bit signal cannot be used, thus wasting the 8-bit signal. Therefore, a relationship between the original density and the image signal is as represented by a solid line A in FIG. 8.
The characteristics of a record head may not, however, have an ability of reproducing a density equal to that in printing. For example, in a record head of an ink-jet printer, the record density depends on the injection amount and the dye concentration of an ink. The ink injection amount is limited by the resolution and absorption properties of recording paper and the like, and the dye concentration of the ink is limited by injection stability and the like. In this sense, neither a large ink injection amount nor a high recording density can be set. In a practical application, a maximum optical density is about 1.5. Therefore, a relationship between the image signal and the recording density is as represented by a solid line B in FIG. 8.
Judging from the above consideration, a relationship between the original density and the output density is as represented by a solid line C in FIG. 8. However, if the maximum density is 1.4 or more, a sufficiently beautiful image can be obtained as a color copy. No problem occurs in copying an image from a normal original. When a copy output is used as an original and copying is performed (i.e., multi-generation copying is performed), a maximum output density is given as follows because the maximum density of the original is 1.5: ##EQU3## The resultant image has poor readability.
When a print original is to be copied, it is known that gradation of a high-density portion is insufficient if an image signal for a density of 1.8 or more is set to be a maximum value. When an image signal for a density of 1.8 is set to be a maximum value, and the maximum value of the recording density is less than 1.8, the repetition density is lowered.
Color reproducibility in a digital color copying apparatus shown in FIG. 2 is greatly influenced by the values of the masking parameters a.sub.11 to a.sub.44. The masking parameters a.sub.11 to a.sub.44 are generally determined by the following method.
Assume that originals having n colors selected from a color space without any discrimination are given as X.sub.1, . . . , X.sub.n. Chromaticity values of these originals are given as (L*.sub.X1,a*.sub.X1,b*.sub.X1), . . . (L*.sub.Xn,a*.sub.Xn,b*.sub.Xn). Assume that these originals are copied to obtain color outputs Y.sub.1, . . . , Y.sub.n, and that chromaticity values of these outputs are represented by (L*.sub.Y1,a*.sub.Y1,b*.sub.Y1), . . . , (L*.sub.Yn,a*.sub.Yn,b*.sub.Yn). A sum of the squares of the color differences between the input and output images is represented by equation (4) below: ##EQU4## The masking parameters are calculated to minimize the sum E.sub.Y-X, which is then defined as an optimal solution.
The parameters thus obtained allow good average color reproduction throughout the color space.
It is, however, impossible to nullify a color difference (difference in color) between an original and its copy even if these parameters are used. It is inevitable that colors change to some extent upon copying processing.
The parameters obtained in the above calculation are enough to obtain satisfactory color reproduction when a normal original is to be copied to obtain a color image. However, a problem is posed in multi-generation copying wherein copying is performed using a copy output as the original.
If a copy output obtained by copying a first generation copy output (Y.sub.1, . . . Y.sub.n) as an original, i.e., a second generation copy output is given as (Y.sub.1 ', . . . Y.sub.n '), a color difference between the original and the second generation copy output is larger than that between the original and the first generation copy output. That is, the color differences between the original and multi-generation copy outputs increase as the number of generations increases. With the recent widespread use of full-color copying machines, multi-generation copying is more frequently performed, thus posing a serious problem.
A parameter determining technique is disclosed in U.S. Pat. No. 5,142,356 wherein a reference image output from a printer is read from a reader and is compared with reference data stored in a memory, thereby determining color processing parameters.
This technique, however, cannot solve the above serious problem posed by multi generation copying.